Practical Acceleration Strategies for the Predictive Visualization of Fading Phenomena

Abstract

Appearance changes caused by light exposure provide important cues that impart a sense of realism to a computer-generated scene. For instance, a carpet may fade or wood may turn yellow over time as a result of many years of light exposure. In this paper, we analyse the key performance and accuracy trade-offs associated with the physically-based simulation of these phenomena. This analysis may be used to guide the selection of simulation parameters in order to achieve optimal color-accuracy and minimize runtime. We also propose a practical method to enable the predictive visualization of these phenomena within applications requiring interactive rates with minimal loss of accuracy. The effectiveness of the proposed techniques is demonstrated through simulations and image sequences depicting fading and yellowing caused by several years of exposure to light.